The Problem with Web Annotations

Written July 25. 1999, at 01:00 GMT.

Red Herring's Attack of the viral Web notes briefly discusses the web's latest trend; Web Annotation software, how there is a growing number of "trendy" annotators on the market and how none of them is compatible with the others.

Also, My friend Bjarni and I recently discussed how badly current implementations of web-annotations scale. They all run into walls wrt usefulness when the number of users becomes too great. The noise level becomes too great.

Bjarni has written down some ideas over at his weblog, on how he thinks IRC-like channels and moderators might be needed.

The fundamental problem I see with all the new and trendy annotation apps so far is that they all lack focus. They all promise to turn the whole internet into one tidy discussion group. None of them take into account that too many users make too much noise. People work best in smaller groups that complement their views and purposes.

I believe we will see several different uses of annotations in the future. Some might be:

• Unfocused graffiti chat amongst friends. Sort of like an extension of the buddy list theme introduced by ICQ and AOL-Messanger. Possibly in the form of instant chat-messages that are "left behind" linked to the page they're refferring to.
• Semi-focused community based discussions Each using it's own Crit-like server and database for comments. For instance a site like Slashdot, could expand its discussion-boards by allowing logged-in members to read and write linked annotations. This could even include some level of moderating.
• Fine-grained linking betweeen published documents. This makes it possible for search engines to trawl for documents that address specific parts of other documents. The fact that the documents would have to be 'published' on the web first, should reduce the noise-level. This functionality is currently supported (optionally) by CritSuite.

This fine-grained linking thing is the core element of annotations. What We need is a standardized (html and http compatible) way to make fine-grained links to documents. Then we can make all sorts of different user-agents that store and search for those links and serve them to the user in a variety of ways. The linking mechanism used by Crit, may be a usable candidate.

More like this: English Entries, .